Which of the following is an example of an industry-level opposed to a firm-level factor?

Numerous studies in the past have provided overwhelming evidence that firms engage in R&D activity to gain sustained competitive advantages and market power (Hart, 1995; Barney, 2000; Rothaermel, 2013). There is also substantial evidence that higher research activity within a firm has a positive effect on its growth (Griliches et al., 1991; Griliches, 1995), as well as on its market valuation (Hall, 2002).1 More recently, some studies have shown strong links between R&D and institutions, and their joint effect on the economic growth of countries. For instance, Coe et al. (2009) argue that countries with a strong patent protection or legal system and where the local conditions permit ease of doing business, benefit more from domestic R&D and international R&D spillovers. Similarly, previous research shows that high-income countries contribute most towards world technological progress. In 1995, the largest seven industrialised (G7) countries in the world captured 84% of the world's R&D expenditure (Keller, 2004). More recent figures show that the high-income countries still dominate global R&D expenditures, which was at 79.7% in 2007 and a slightly lower share of 69% in 2013 (UNESCO, 2015).2 Thus, the extant literature provides strong evidence of the importance of R&D and its relationship with institutional elements of those countries.

However, to what extent does the above evidence suggest that country-level factors, such as income, trade or income growth rates matter when it comes to R&D investment by firms? To date, we find very little evidence in the extant literature that empirically investigates the relative importance of firm-level vis-à-vis country-level determinants of R&D. Recent findings show that China's share of research expenditure has almost doubled from 10.2% in 2007 to 19.6% in 2013. Moreover, 84% of this research expenditure in China is experimental compared to only 64.2% in the US (UNESCO, 2015). South Korea, another transitional economy, is among the top five countries in the world in research expenditure in 2013 ($69 billion) Thus, it appears that the income status of a country may not be the sole driver of R&D intensity and rigorous empirical testing of this hypothesis is required before any meaningful conclusion can be reached.

R&D expenditure is often considered as a risky investment by firms (Bhagat & Welch, 1995; Kothari et al., 2002; Coles et al., 2006). There is no certainty that it will always offer a high return and offset the associated costs. Therefore, firms pay attention to risks associated with the host country's business environments and institutional characteristics. Effective country-level governance characteristics, such as a robust legal system or a better financial system, provide better investor protection, which increase the market valuation of R&D projects of firms (Pindado et al., 2015) The importance of government policy or geographical location in R&D is highlighted by Porter (1990, 1994, 1998). Porter and Porter (1998) argues that the efficiency and competitiveness of domestic and foreign firms are strongly influenced by the quality of institutions in the country where these firms are domiciled. For instance, if transaction costs are very high, there is a high level of red tape, low quality of transport and research infrastructure, low protection of intellectual property rights, and other similar attributes are visible. Thus, these firms are unable to compete on high service strategies and end up providing low customer satisfaction.

Other studies that show the importance of institutional quality or intellectual property rights in determining the level of competition and R&D are Guellec and Potterie (2004), Acemoglu and Akcigit (2012) and Spulber (2013). While Guellec and Potterie (2004) show that institutional quality plays a key role in determining the effect of R&D on the income growth of countries, Acemoglu and Akcigit (2012) argue that policies on intellectual property rights (IPR) in a country should provide greater protection to the technology leaders (innovative firms) to achieve higher growth. Moreover, Spulber (2013) finds that antitrust policies and IPR policies are complementary in providing incentives among firms to innovate. Thus, it appears that country-level institutional characteristics play an important role in a firm's R&D investment decisions.

Alternatively, with a higher level of international capital flows among countries in the form of international trade and foreign direct investment (FDI), the role of country-level characteristics may diminish over time and the outcome of an R&D investment is then dominated by firm- or industry-specific factors. For instance, internationally product diversified firms may benefit more from R&D intensity than non-diversified firms in the same industry group (Hitt et al., 1997). This is mainly because by integrating across country borders and standardising their products, these internationally diversified firms minimise their opportunity cost of scarce resources, such as R&D, and enjoy both economies of scale and economies of scope (Kochhar & Hitt, 1995). Moreover, technological opportunity and appropriability at the firm level differ greatly across industries and time (Gilbert, 2006). For instance, Basant and Fikkert (1996) show that for Indian manufacturing firms, the returns to foreign technology and in-house R&D varies greatly between the non-scientific firms and the scientific firms in the early 1990s before the trade reforms took place in India.

Then, to what extent do firm, industry- and country-level determinants drive firm-level R&D? The extant literature is silent on this question. This study uses a comprehensive dataset covering 94,178 firm-year observations from 42 countries spanning 1981–2013 and examines this issue by capturing multiple dimensions of market and institutional characteristics. Our findings suggest that the firm, industry- and country-level determinants jointly have maximum explanatory power. However, firm- and industry-level determinants have higher explanatory power than country-level determinants. When R&D is scaled by the total assets, firm- and industry-level determinants jointly explain 45.5% of the variations in the sample. However, country-level determinants explain only 7.5% of the variations in the sample. Likewise, when R&D is scaled by the total number of employees, we find firm- and industry-level determinants jointly explain 48.3% of the variations in the sample. On the contrary, country-level determinants explain only 7.1% of the variations in the sample. When we include firm, industry and country-level determinants in a single regression, they jointly explain between 46% (when R&D is scaled by total assets) and 49% (when R&D is scaled by the total number of employees) of the variations in the sample. Since, the adjusted-R2 values after including all three types of determinants, i.e., firm, industry and country-level determinants, closely match with the adjusted-R2 values explained by firm- and industry-level determinants, the latter dominate the variations in the sample. After bifurcating our sample based on weaker and stronger institutional scores related to R&D, firm- and industry-level determinants still dominate. Within each category, country-level determinants explain higher variations in the sample with weaker institutional scores. The effect is also similar for different industry-level splits. For instance, we find that firm-level determinants have a higher explanatory power for the healthcare (39.7%), telecommunications (21.6%), industrial (27%) and technology (21.6%) sectors. These results are robust to adopting alternative empirical specifications, such as R&D stock measures, and correcting for non-zero R&D values.

The study contributes to the literature in two significant ways: first, to our knowledge, this is the first study in the R&D literature to empirically test the relative importance of country-versus firm- and industry-level determinants, capturing both cross-sectional and time series variations in the sample. Although this issue has been considered in the past in the context of firm-level corporate decision-making outcomes, such as corporate governance undertakings, capital structure, environmentally sustainable practices and leverage, similar approaches are yet to be undertaken in the context of R&D.3 This is a relatively new area of research in the field of economics and finance, and thus remained under-researched until now. However, the topic is becoming increasingly popular and requires more attention in the era of financial globalisation, which allows free movement of capital flows among countries. For instance, when a multinational firm is deciding whether to invest in R&D in a developing country, it is increasingly important to understand the existing market dynamics. If findings suggest that the country-level determinants have a very high explanatory power, then depending on the choice of country for R&D investment, firms need to factor in all costs associated with the provision of infrastructure, the protection of property rights and other transaction costs, which would have been otherwise provided by the host country. Hence, to minimise costs, the firm can channel resources towards R&D related to FDI in host countries that demonstrate strong economic growth and institutional characteristics. Alternatively, if the results suggest that country-level determinants have a low explanatory power and most of the variations in the determinants of R&D are driven by firm and industry-level characteristics, then the choice of a host country is probably less important for the firm to make any new investment in R&D. In this scenario, to protect itself from unstable local business environments of the host country, it is essential for the firm to have a robust internal organisational structure. By undertaking prudent corporate governance practices, the firm can ensure that investment in R&D turns out to be value-enhancing.

Second, recent evidence suggests that there exists a geographical bias in research fields such as economics and finance, where an increasing number of empirical works are published on topics related to the US market in contrast to a broad range of international markets (Das et al., 2013; Karolyi, 2016). However, since the 1990s the impact of research papers written in the areas of international economics and finance is higher and they also received a higher number of citations (Card & DellaVigna, 2013; Karolyi, 2016). Karolyi (2016) argues that an important factor limiting research works in a multi-country setting (in addition to countries such as the US, China and some European countries) is the availability of a comprehensive dataset, which are both expensive and only recently available to many universities across the world.4 Our study clearly overcomes this limitation and considers a unique database, consisting of firm-level data from 42 countries, covering both surviving and delisted firms. Yet another important contribution of this study is the focus on the effect of institutional differences and industry differences among countries on the relationship between firm- and industry-versus country-level determinants of R&D. Thus, findings from this study will help researchers to form a better understanding about the relative importance of firm- and industry-versus country-level determinants of R&D intensity in a multi-country setting.

The rest of the paper is organised as follows: Section 2 discusses the different firm-level and country-level determinants of R&D citing the relevant literature. Data and methodology are presented in Section 3. In Section 4 we present the empirical results and their implications. Finally, Section 5 summarises the research findings and conclusions.

Firm-level determinants of R&D

Before we empirically test the relative importance of country-versus firm-level determinants, it is important to study each firm-level determinant of R&D based on the relevant theoretical and empirical literature. The relevant set of firm-level determinants will lead to a less chance of model misspecification or biased results due to omitted variables. We refer to the extant literature and use the following firm-level determinants: log of total assets, price to book value ratio (PTBV), cash

Sample data

Our sample consists of 94,178 firm-year observations and 650 country-year observations from 42 countries. Our sample is devoid of survivorship bias, as firms are added and delisted over the sample period. The sample includes high income, middle income and low-income economies and covers a period of 1981–2013, providing a robust panel data.

The list of countries and a summary of the sample is provided in Table 1. It is evident from Table 1 that the sample is not evenly distributed, with some

Baseline results

We present the baseline results in Table 4. Models 1 to 5 are estimated using three different measures of R&D intensity: R&D expenditure scaled by the total assets (Panel A), R&D expenditure scaled by the total sales (Panel B) and R&D expenditure scaled by the number of employees (Panel C), respectively. Under each panel, five different models are estimated to better understand the effects of firm-versus-country level determinants on R&D intensity. While in model 1, we include only

Conclusion

To the best of our knowledge, this is the first paper to test the role of firm- and industry-level determinants vis-à-vis country-level determinants of R&D in a multi-country setting. We pay particular attention to the institutional and industrial differences across countries and also to different measures of R&D intensity while testing the relationship between firm, industry- and country-level determinants of R&D.

Our findings suggest that firm, industry- and country-level determinants of R&D

Credit author statement

Banerjee, Rajabrata: writing- original draft preparation, investigation, reviewing, editing.

Gupta, Kartick: conceptualisation, methodology, empirical analysis, editing.

Acknowledgements

The authors are grateful to Ms. Ashleigh Campbell of the UniSA Business School, University of South Australia and Semiota editorial service for their editorial comments on earlier drafts of this paper. Comments received from colleagues at the UniSA Business School are also gratefully acknowledged.

Cited by (1)

Recommended articles (6)